The Impact of State-owned Urban Land Leasing on Intention to Reside of Migrant Labors

— Based on the data of the 74310 individual migrant labors from 250 cities that distribute uniformly in China, taking the fiscal stress of the local governments as the instrumental variable due to endogeneity, the IV-probit models indicate that: the area of state-owned urban land leasing promotes the overall intention to reside of migrant labors. Besides, the use structure of urban land leasing matters as well. Land leasing for public service effects the intention to reside positively while land leasing for house building and the secondary industry restrain the intention through mediators like education, house price and industry scale. State-owned urban land leasing for the tertiary industry is proven uncorrelated with the intention to reside, however. It is advised that the local governments ought to choose the urban land leasing structure with larger proportion of land for public service and less proportion of residential and industrial land to strengthen the intention to reside of migrant labors.


Introduction
The mobile population have been one of the major issues worldwide [1] and so are they in China. The Mobile Population Development Report of China 2018 indicates that accounting for nearly one fifth of the country's whole population, the size of the mobile population in China reached 245 million in 2017. At the mean time, the mobile population size in China shrank from 253.53 million to 245 million during 2015-2017 constantly, which implies the growing intention to reside of the mobile population.
As a large proportion in the mobile population, migrant labors are of vital importance for urban development. Researchers have found that migrant labors can be the supplement for the urban labor market based on the evidence in Shanghai (Wang and Shen, 2001) [2] . The research based on Shanghai also proves that the output elasticity of migrant labors is higher than the residential labors (Yang and Gao, 2012) [3] . City administrators have to rely on migrant labors due to the lower birth rate in recent years, to hedge the aggravating negative effect brought about by population aging (Yuan and Jin, 2020) [4] . High-quality talents among migrant labors are also of great benefit for independent innovation and industrial upgrading in urban area (Cui and Gao, 2019) [5] . In fact, over 50 large or medium cities in China have implemented policies to make migrant labors reside in 2019.
State-owned land leasing is one of the most active fiscal activities for local governments in China, which can be the base of city construction, industry development and public service supply according to existing researches [6] . The research based on the panel data of 36 cities located in the middle reaches of the Yangtze River finds that urban land leasing promotes construction of infrastructure (Li, Gu and Zhu, 2020) [7] . Marketized urban land leasing promotes the green total factor productivity and industrial structure supererogation (Jiang, Lu and Gong, 2019) [8] . Meanwhile, the residing decision of migrant labors lies on the utility of urban life (Bondy, 2020) [9] , which is potentially correlated with urban land leasing and the following-up construction. However, this potential correlation has not been of much mention among researches so far.
Based on IV-probit model and statistics from the China Migrant Population Dynamic Monitory Statistics, according to which "reside" is defined as "live for 5 years or longer", the China City Statistical Yearbook and the China Land Resource Yearbook, the conclusion is that state-owned urban land leasing promotes the intention to reside of migrant labors through several mechanisms including house price, public service and industrial scale.

The explained variable
The explained variable is the intention of an individual migrant labor to reside in the certain urban area for 5 years or longer according to the survey in 2016, and "1" is assigned to "yes", "0" to "no".

The explanatory variables
The core explanatory variable is the total area of stateowned urban land leasing during 2012-2014 of each city considering the interval between land leasing and the utility of the following-up construction.
The control variables include the average house price, the GDP per capita, the proportion of the secondary industry, the proportion of the tertiary industry, the population density, the average wage, the teachers per primary and middle school student, the doctors per capita and other indicators of each city in 2015 to control the proven effects from city economy and employment (Zhang and Yang, 2020) [10] , public service (Xia and Lu, 2015) [11] and other factors on the intention to reside.

Endogeneity and Instrumental Variable
Considering endogeneity in the probit model, the budgetary fiscal stress of local governments is selected as an instrumental variable(IV), which equals to budgetary expenditure minus budgetary revenue. Existing research shows that state-owned land leasing is a common method for local governments to release their fiscal stress (Zheng, 2020), and the weakiv test aims to prove the correlation between IV and core explanatory variable.
The exclusion restriction is vital for an IV, which means IV effects the explained variable only through the endogenous variable. On one hand, the IV is strongly correlated with budgetary expenditure and budgetary revenue which is included in model residuals. However, budgetary revenue do not have much correlation with the intention to reside because of the little difference in taxation among local governments in China, which means that as tax burden is the most influential proportion in budgetary revenue for individual migrant labor, it ranges slightly among different cities for the same person, turning out to be a trivial factor on intention to reside of migrant labors. Budgetary expenditure does not have much correlation with the intention to reside, neither. Expenditure will not necessarily turn into utility of migrant labors with an inappropriate expenditure structure. On the other hand, the IV does not have much correlation with other control variables. Fiscal stress reveals the necessity for more revenue through land leasing while it does not effect the level of city development indicators so that the IV is not correlated with the control variables, neither. In conclusion, the IV does not violate the exclusion restriction.

The descriptive statistics
The descriptive statistics of variables are as table I:

Introduction of IV-probit model
The probit model is a typical binary selection model. Given that: Y is an incident with binary results and x is a factor that effects y. β is the coefficient.
is a standard normal cumulative distribution function (CDF), then: Model (3) is the general form of probit model. Φ means standard normal distribution. For probit model with endogeneity, the instrumental variable(IV) is necessary to make the estimate consistent. IV-probit model is the probit model with IV. The usual method to construct an IV-probit model is the control function approach. Compared with the multiple linear regression, the IV-probit model is more accurate in depicting binary selection and more effective in avoiding endogeneity.

Impact of State-owned Urban Land Leasing
State-owned urban land leasing is the base of the following-up city construction which may promotes utility of migrant labors and their intention to reside eventually. Based on this logical chain, the first hypothesis is as follows: Hypothesis 1: State-owned urban land leasing effects the intention to reside of migrant labors positively.
Model is set up to test the hypothesis as followed: In (4)

The Overall Impact of State-owned Urban Land Leasing
The result of regression based on (4) (4) is the fiscal stress of the local government in the certain city. Result of Wald test of exogeneity shows that we can view endogeneity exists in the model at the significance level of 5% so that the IV is necessary. On the other hand, the result of Anderson-Rubin test indicates that it's reasonable to deem there is no weak IV problem in the model at the significance level of 5%, which means the IV is valid in the model.
The control variable gdp per capita reflects the overall level of city economic development, which effects the intention to reside positively in accordance with the existing researches. Lower city house price is able to cut down the living cost of migrant labors and higher wage raises the level of utility, which effect the intention positively. Public services like education, healthcare, transportation offer additional utility to migrant labors to strengthen their intention to reside.
With the variables mentioned above controlled, the core explanatory variable, area of urban land leasing is positively correlated with the intention to reside at the significance level of 1%. In order to identify the potential causality between urban land leasing and intention to reside, the test of mechanisms where the causality lies is necessary.

Test of Mechanisms
Based on (5), (6), (7), (8), the results of regression are as According to the results of Wald test of exogeneity and Anderson-rubin test, it's reasonable to view that endogeneity exists and weak IV problem does not exist at the significance level of 5% in (5). (6), (7), (8), which proves the validity of the instrumental variables.
Besides area of urban land leasing for industry, commerce, house and public service, all other explanatory variables are all controlled in (5), (6), (7), (8) as in (4), too. As the length limits, their coefficients and standard deviations are not showed in the table. Readers who are interested may ask for these from the author.
In (5), the sample consists of individuals who work in the secondary industry since they are most sensitive towards the alteration of this industry logically, which makes, if it exists, the potential causality easier to identify. The result indicates that the cross term ind_sec effects the intention negatively, which means along with the area of industrial land leasing expanding and the secondary industry being enlarged, the intention to reside of migrant labors goes down correspondingly. Noted that the proportion of the secondary industry has fallen form 46.80% to 40.50% during 2010-2017, the adjustment of industrial structure indicates that the external diseconomies of scale exists in the secondary industry in China. As the secondary industry expands due to land leasing, the marginal benefits shrink because of the extra costs from the oversized industry, which is transmitted to the intention of migrant labors through utility.
In (6), the sample consists of individuals who work in the tertiary industry for the similar reason, while the result indicates that the cross term com_ter is not correlated with the intention. Since the proportion of the tertiary industry has not fallen down in recent years, it's reasonable to view that the diseconomies of scale does not exist in the tertiary industry for now. In that case, it's hard to affirm the effect of land leasing on the intention to reside transmitted through utility.
In (7), the sample consists of the individuals that haven't purchase houses in the certain city and are more sensitive to the house price, which helps to make the result easier to observe. The result shows that the cross term hou_hou effects the intention to reside positively at the significance level of 1%. Current researches have proven that urban land leasing makes the house price increase significantly (Zhang, 2019) [12] , which burdens the migrant labors who have not purchased houses with a higher expected living cost that restrains the intention to reside.
In (8), the sample consists of the individuals that are willing to register their residence in the certain city. Logically, since public service is closely linked to the registration of residence in China, people who are willing to register the residence care more about the public service in the city, which makes the potential causality more obvious. The result indicates that the cross term pub_tea effects the intention to reside positively at the significance level of 1%. The expansion of the area of public service land leasing boosts the local education, which makes the migrant labors more likely to reside for the better chance that the children they have or will have can receive high quality education.
The results of the IV-probit regression indicate that the hypotheses mentioned above are basically tenable. Among the hypotheses, 2b and 2c are tenable according to the results of regression, while 2a is half tenable since there no correlation between the commerce land leasing and intention to reside. Hypothesis 1 is tenable based on 2a, 2b and 2c.

The Conclusions
Based on the sample consists of 74310 individual migrant labors from 31 provinces in China, the empirical analysis indicates that: firstly, urban land leasing promotes the overall intention to reside of migrant labors positively through several mechanisms. Secondly, public service land leasing effects the intention to reside positively through enriching the local educational resources. Finally, industrial and residential land leasing effects the intention to reside negatively when they enlarge the scale of secondary industry where the diseconomies of scale exists or increase the house price that brings out extra living cost.
It is advised that the local governments ought to pay more attention to the use structure of urban land leasing. To strengthen the intention to reside of migrant labors, larger area of public service land leasing and less area of residential land leasing are desirable. Besides, less industrial land leasing is conductive to promoting the intention to reside in accordance with the current trend of industrial upgrading. The urban land leasing with the well-tuned use structure boosts the overall intention to reside of migrant labors.